Advances in artificial intelligence mean it is becoming easier to create highly realistic deepfake videos, which can appear to show someone doing or saying something they did not in fact do or say. Deepfakes may present a threat to individuals and society: for example, deepfakes could be used to influence elections by discrediting political opponents. Psychological research suggests that people cannot reliably identify deepfakes and thus can be influenced by their content. However, little is yet known about the potential impact of a deepfake video which has been explicitly identified and flagged as fake. We explored this issue with two preregistered studies in which participants were shown a bespoke deepfake video of a man appearing to admit to committing a crime. Participants were then asked questions about the man’s guilt, to test how they had been influenced by the video’s content. We found that most participants relied on the content of a deepfake video, even when they had been explicitly warned beforehand that it was fake. This result was observed even with participants who indicated that they believed the warning and knew the video to be fake. We also found this specific warning to be no more effective than a generic warning about the existence of deepfake videos. Our findings suggest that identifying and flagging deepfake videos will not entirely negate their influence. This has implications for legislators, policy makers, and regulators of social media platforms and online news.
No takes yet. Share an insight, caveat, or question.
Clark et al. (2024) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: